Comparative Analysis of CNN and Transformer models for image classification on Intel Image dataset
Year: 2025
Type: conference-paper
Abstract: This comparative study investigates the use of six neural network architectures-ResNet50, VGG16, EfficientNetB0, Vision Transformer, Swin Transformer and DenseNet -for classifying images in the Intel Image dataset.The research conducted tends to discover the disadvantages as well as advantages of each model in terms of accuracy, computational efficiency, and versatility.By implementing and fine-tuning these architectures on the chosen dataset, we assess their ability to categorize various image types.This study provides insights of the balance in between model complexity and as well as performance, offering valuable guidance for researchers and professionals in selecting appropriate neural network architectures for diverse image classification tasks.
Authors: Shreedatta Sawant, Ashish Narvekar, Aditya Pednekar, Dinesh Devalie Naik, Dhruv Malvankar
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4411094198